Distribution of Personal Income in *: Evidence from a Microeconomic Survey

Eiji Tajika Izumi Furutani

Abstract An important feature of the Japanese personal- system is that the role of taxation has been confined to collecting revenues and very few tax credits have been introduced as a means of welfare policy. The personal income tax has instead achieved progressivity by applying various deductions to . This was made possible by sharply increasing the marginal for people in higher income brackets. The paper shows who pays the tax using a microeconomic survey, and examines the effects on average and marginal tax rates by reforming the current Japanese system to a linear income tax, which is constructed by a basic transfer and a uniform marginal tax rate.

Keywords: average tax rate, marginal tax rate, taxable income, tax credits, linear income tax

1. Introduction There are, thus, two characteristics of the Japa- Personal income tax is collected from taxing labor and capital nese personal income tax. First, capital income is sepa- income. In Japan, taxing both types of income comprehen- rated from labor income and it is often taxed at lower rates sively has met with many difficult problems, and over the than labor income (interest and capital gains were at one years capital income taxation has been set separately from time almost tax free). Second, the personal income tax does labor income taxation. Moreover, capital income tax, no not take into account mitigating the poverty by distribut- matter whether it is levied on interest, dividends or capital ing credits. As a consequence of this background, the gains, is mostly a withholding tax and the tax rate for some personal income tax in Japan has been designed to raise capital income (say, interest and the capital gains realized substantial revenue from labor income and the way it is from selling equities) is uniform. collected is progressive. Labor income taxation in Japan Another feature of the Japanese personal income may be claimed to be self-contained; from within labor tax system is that taxation is considered to be one thing income alone, the tax has to collect the required revenue and welfare another. That is, the Japanese Ministry of Fi- with a reasonable degree of progressivity. nance has been very much concerned about losing rev- The purpose of this paper is to unveil the conse- enues and has been very reluctant to give credits from quences of the intent of those who have designed the what it collects from the people. The Ministry would con- Japanese personal income tax. We will concentrate our tend that the role of taxation is to collect revenues and examination on earned income: that is, the income from what needs to be distributed to the poor and other grant- wages and salaries, and show how this income is taxed. deserving people should be managed by welfare policy, We classify the income into ten brackets and present the which in Japan is delegated to other ministries. Combined following information for each of them: average and mar- so far with the paucity of single-mothers and other welfare ginal tax rates, progressivity of tax, and the extent to which dependency problems, which have triggered various types the tax base is eroded by various income deductions. of earned-income tax credits in the United States and Eu- We show that the tax burden has in fact been very ropean countries, the Japanese tax system has not ap- progressive and it has been shifted to the highest income pealed to tax credits1. group. The extent of the tax base erosion by deductions is

S 17 also shown to be very significant, but it has big variance and marginal tax rates in this paper, social security contri- over individuals as well. On the basis of these findings, we butions are also taken into consideration, for they are pro- introduce tax credits to the Japanese personal income tax portional to earnings. As in other countries, social secu- and show how the tax burden of the rich, especially their rity contributions of employees are split equally between marginal tax rate, will change. We then proceed with this employees and employers. However, since only the em- experiment by replacing the current personal income tax ployees’ contribution is reported in the data-base we use, system with the linear income tax system, which consists we set the social security contribution equal to the of a basic transfer and a uniform marginal tax rate. employee’s part. The distributional effects of the new tax will partly As was alluded to in section I, our estimation was depend on how people react to it by changing their supply done using Basic Survey on People’s Life. The data we of labor, which is an incentive aspect of the taxation. Bar- used for estimating individual tax burden was taken from a ring this, the results depend on the size of the lump-sum survey conducted in 1998 and contain information on in- transfer that is granted. Setting it at some portions of the come and tax for the year 1997. Amongst the total of 90,059 average tax collected under the current tax system, say people surveyed there are 54,642 people with positive in- 20% and 50%, we find the marginal tax rates of the top two come. We classify income into earned income, self-em- brackets decrease. The other side of the coin is that the ployed income and pensions. For our study here we have burden of the tax is shifted to lower income groups, but, chosen those who receive only earned income of the three since it is spread over many people, the increase of the types of income. The number of people satisfying this average tax burden is not very large. selection criterion is 32,672 with average income of Our study uses Basic Survey on People’s Life of 4,407,000 yen (JPY, hereinafter). the year 1998, complied by the Ministry of Health, Labor The reason for focusing our attention on earned and Welfare. This is a collection of household data relat- income is that the size of samples receiving this income is ing to the health, income, and savings status of individu- far larger than the numbers of the self-employed and pen- als and families. Surveys are conducted throughout the sion-receivers, which are 2,467 and 13,101 respectively. entire nation every three years by the Ministry. Specific Moreover, since the purpose of this study is to show evi- features of the survey will be explained in Section II; it dence of tax burden, we thought it better to concentrate suffices here to say that we are using income and saving on one type of income; comparison of tax burden among data of a microeconomic survey which contain individual different sources of income would be better dealt with in a income tax (both central and local) and social security con- separate paper. tributions. Table 1 shows the average and marginal tax rates The paper proceeds as follows. The next section for the ten income classes with overall average and the explains the outline of the survey that we use for our study total number of people at the bottom. The two tax rates are and reports the basic statistics about the progressivity of estimated first for the personal income tax levied by the the Japanese personal income tax. The third section dis- central government, second for the sum of the central and cusses the effects of deductions on the taxable income. local personal , and third for the sum including social The fourth section will show the effects of the reform that security contributions. There are a couple of facts to be replaces the current Japanese tax system with the linear noted. As for average tax rates, they are increasing along income tax on average and marginal tax rates. The last with the increase of income. It is somewhat surprising to section concludes and discusses other important aspects find that the absolute magnitude of personal income tax— of the Japanese personal income tax for future study. i.e., central and local personal income tax combined—is not very high; it starts at almost zero in the lowest bracket to 0.131 at the top with overall average 0.075. 2.Progressivity of Japanese Personal In- However, it should also be noted that when social come Tax security contributions are added to personal income taxes, average burden on earned income is not small at all; except We classify people receiving earned income into deciles for the top bracket, the sum of the three burdens are more and show how progressive their personal income tax bur- than twice the sum of the personal income taxes. Since dens are. Both the central and local governments collect only the employee’s contribution is taken into consider- personal income tax in Japan. When calculating average ation here, the combined burden of tax and social security

S 18 is even larger. the index of progressivity of tax burden. The same mea- Marginal tax rates are calculated by dividing the sure applies to local personal tax and to social security aggregate increment of tax and social security contribu- contributions. This index implies that the burden of tax tions by that of income for each income class. They start and social security contributions is progressive when it from the second lowest bracket, for the income class be- exceeds unity in the sense that the share of burden is more fore the lowest—in fact, this should be the minimum tax- than proportional to income share. able income—depends on individual circumstances and Table 2 reports the income and revenue (burden) is therefore not available. shares for each of the ten brackets. The last three columns Marginal tax rates are not increasing very much are the measure of progressivity and they show very clearly and there are even some declines up to income class 7, the fact of tax and social security burden in Japan. The which has a marginal tax rate slightly above the average. burden of the central personal income tax becomes only However, for income class 8 and above, marginal tax rates progressive in the top income bracket and the progressivity start to increase sharply; the marginal tax rates of the cen- index jumps from about unity in the ninth bracket to 1.7 in tral personal income tax are 0.054, 0.080 and 0.138 respec- the top bracket. Almost the same results are observed when tively for the eighth to the top income brackets; and a central and local personal income taxes are combined. The similar observation applies to the marginal tax rates incor- degree of progressivity is reduced when social security porating the other two burdens. contributions are combined with the tax burden, but the We turn to the progressivity of the tax and social intrinsic nature of progressivity of the burdens is un- security contributions. Here, we define the progressivity changed. of tax burden for an income class as the ratio of its revenue share to income share. Results are shown in Table 2. In the table, the income share of the lowest income class is 1.5% 3. Deductions and the Erosion of Tax Base and the share of the central personal income tax paid by A characteristic of the Japanese personal income tax is this class is 0.1%. In this case, 0.07 (0.1 divided by 1.5) is that various income deductions (deductions in short, here-

Table 1: Average and Marginal Tax Rates by income class JPY 1,000 Income Number of Average class people Income Average tax rate Marginal tax rate Central Local Central Local Central and and Social Central and and Social Central Local Security Central Local Security 1 3,267 669 0.003 0.005 0.034 - - - 2 3,267 1,253 0.013 0.020 0.081 0.025 0.037 0.135 3 3,267 1,972 0.025 0.036 0.115 0.045 0.063 0.173 4 3,267 2,625 0.028 0.042 0.123 0.039 0.060 0.149 5 3,267 3,271 0.032 0.048 0.127 0.047 0.071 0.143 6 3,267 4,011 0.033 0.050 0.134 0.039 0.059 0.165 7 3,267 4,874 0.035 0.053 0.134 0.041 0.071 0.131 8 3,267 5,979 0.038 0.061 0.145 0.054 0.095 0.194 9 3,267 7,518 0.047 0.077 0.159 0.080 0.137 0.215 10 3,273 11,888 0.080 0.131 0.198 0.138 0.223 0.264 Overall 32,676 4,407 0.047 0.075 0.152 - - - Notes: (1) The year of the estimation is the calendar year of 1997. (2) JPY: . (3) “Central” and “Local” stand respectively for centrally and locally collected personal income tax. And “Social Security” is the sum of total social security contributions.

S 19 inafter) are granted to achieve distributional equity. This by deductions. Here, the deduction rate is defined as the has resulted in erosion of taxable income. An important proportion of the estimated amount of deductions to in- aspect of this problem is that deductions depend very come. A problem for estimating this rate is that deductions much on the family and other individual circumstances: are not reported in the survey. There are two ways for that is, people receiving the same income may face differ- obtaining the amount of deductions: the first is to examine ent tax rates due to their individual situations, which may the family circumstances that are shown in the survey and not reflect their earning capacity. This is but one way of to calculate internally individual deductions by using vari- looking at the issues of deductions, and there certainly ous tax codes; the second is to infer the taxable income should be another approach that looks at them more fa- from the tax paid using the tax table and to estimate the vorably. This section seeks to show the evidence of de- amount of deductions as the difference between the in- ductions in Japanese tax system. Observations are re- come and the taxable income thus estimated. stricted to the central personal income tax for the brevity The first method is like itemizing deductions and of presentation. may be better for simulating the change of tax liabilities Table 3 reports how much taxable income is shrunk when some deductions are eliminated or changed. But we

Table 2: Progressivity of personal income tax and social security contributions Income Income Share Revenue share Progressivity of tax class % % and social-security contributions Central Central Central Local and Central Local and and Social and Social Central Local Security Central Local Security 1 1.5 0.1 0.1 0.3 0.07 0.06 0.23 2 2.8 0.8 0.7 1.5 0.28 0.26 0.54 3 4.5 2.3 2.1 3.4 0.52 0.47 0.76 4 6.0 3.6 3.3 4.8 0.60 0.55 0.81 5 7.4 5.0 4.7 6.2 0.68 0.63 0.84 6 9.1 6.4 6.0 8.0 0.70 0.66 0.88 7 11.1 8.1 7.8 9.7 0.73 0.71 0.88 8 13.6 11.0 11.0 12.9 0.81 0.81 0.95 9 17.1 16.8 17.3 17.9 0.99 1.02 1.05 10 27.0 45.8 46.9 35.2 1.70 1.74 1.30 Note: Progressivity of tax burden of an income class is defined by the ratio of its revenue share to income share. For example, the income share of income class 1 is 1.5% and the share of the central government’s personal income tax paid by this class is 0.1%. In this case, 0.07 (0.1 divided by 1.5) is the index of progressivity of tax burden. The same measure applies to local personal tax and to social security contributions.

Table 3: The Deduction Rate (A) Deduction rate by income class (B) Dispersion of deduction rate by income class, % Average The Income Number of Income Deduction Income Class people JPY1000 Rate % Class 0 to 20% 20 to 40% 40 to 60% 60 to 80% 80% over 1 2,268 1,430 70.3 1 2.2 6.0 16.8 39.9 35.0 2 2,268 2,229 64.5 2 2.2 9.4 20.6 51.0 16.8 3 2,268 2,852 64.1 3 1.9 7.5 26.1 47.4 17.2 4 2,268 3,488 62.3 4 3.2 5.0 40.1 35.7 16.0 5 2,268 4,157 61.9 5 3.8 6.8 36.4 34.9 18.1 6 2,268 4,893 62.5 6 2.3 8.5 33.0 39.4 16.8 7 2,268 5,774 60.9 7 2.2 8.7 34.8 41.7 12.6 8 2,268 6,857 59.1 8 1.5 7.6 42.2 39.7 9.0 9 2,268 8,257 54.5 9 1.6 12.1 54.5 27.0 4.8 10 2,277 12,479 44.7 10 5.8 31.0 43.3 17.4 2.5 Overall 22,689 5,245 56.7 Note: The deduction rate is the proportion of estimated amount of deductions to income.

S 20 have to identify the contents of deductions for each indi- to achieve equity in tax burden in a closed arena of tax vidual. The second is more straightforward and gives us without a perspective of distributing from the total amount the total amount of deductions as simply the difference of tax collected to the poor. between the income and the taxable income. We have cho- Deductions are not only very large in comparison sen the second method to estimate deductions. with the income earned, but their variance is significant as Table 3 shows our results. The number of people in well. Table 3 (B) shows the dispersion of the rate of deduc- the table is smaller than that reported in Tables 1 and 2. tions for each income class. The average deduction rate of This is because the data that have missing values in the income class 6 is 62.5%;e however, it varies from 0-20% to amount of tax have to be eliminated (recall that we calcu- “80% over” with the two largest groups, 39.4% and 33.0% late back the taxable income from the amount of tax paid). respectively, in the ranges of 60-80% and 40-60%. More- About 10,000 people of the total of 32,676 now disappear over, 16.8% of those in the average income group have from our study. This has resulted in increasing incomes deducted more than 80% of their income to reach their across all brackets and the average income increases from taxable income. JPY4,407,000 in Tables 1 and 2 to JPY5,245,000 in Table 3. These observations apply to the highest bracket, The results in Table 3 (A) are indeed striking. The too. While its average deduction rate is 44.7% and is still rate of deductions starts with 70.3% at the lowest income very high, it is startling to find that 43.3% and 17.4% of bracket and declines to 44.7% at the highest bracket. This those in this bracket cut their taxable income by 40-60% rate is more than 60% even for the average income: that is, and 60-80% respectively. In sum, the table seems to sug- more than a half of the average income is eroded when gest that granting deductions in the personal income tax calculating the taxable income. This is a consequence of may not be as fair as it intends to be. the Japanese personal income , which has sought Given the results of Table 3, we have simulated the

Table 4: The effects of cutting deduction on (A) The rate of increase of tax revenue by cutting deductions, % The rate of deduction cut Income Class 10% 30% 50% 100% 1 22.7 68.1 113.5 227.1 2 17.3 52.0 86.7 173.4 3 17.3 52.0 86.7 173.3 4 16.4 49.2 82.1 178.2 5 16.4 49.2 82.6 214.1 6 16.8 51.2 91.3 245.9 7 16.3 52.8 100.1 247.1 8 17.2 58.1 108.9 242.0 9 16.5 53.5 93.4 193.5 10 10.4 32.8 56.4 124.7 Overall 14.3 45.1 79.3 178.4 Note: The rate of deduction cut is the proportion of deductions that are slashed to increase the taxable income, i.e., when the rate is 100%, the earned income is identical with the taxable income itself. (B) Dispersion of the rate of increase of tax revenue by income class when deductions are cut by 30%. The rate of increase of tax revenue Income Class 0 to 20% 20 to 40% 40 to 60% 60 to 80% 80% over 1 9.6 13.1 13.8 10.5 53.0 2 12.8 15.7 21.0 21.1 29.4 3 10.6 16.7 33.4 10.7 28.6 4 8.4 26.3 28.6 9.2 27.5 5 10.1 32.0 15.3 12.0 30.6 6 4.3 33.7 18.6 12.4 31.1 7 4.6 15.2 39.6 14.2 26.5 8 3.9 10.9 20.1 38.3 26.8 9 5.3 20.2 22.3 15.7 36.5 10 16.5 39.7 17.6 7.8 18.4

S 21 effects of the increase of tax revenue by cutting deduc- for the people in those brackets enjoy more deductions in tions. Table 4 is the outcome of this experiment. Table 4 proportion to their income. But in every income bracket, (A) shows that when deductions are cut by 10% the over- there is very wide dispersion of the rates of increase of all tax revenue increases 14.3%. When deductions are cut revenue. by 30%, the revenue will be up by 45.1%. When they are Let us take a look at income class 6 again, which is completely removed, total revenue will increase almost an average income group. The overall rate of increase of threefold. tax revenue in this bracket is 51.2%, but for the largest This is merely another side of Table 3 that shows fraction of the people of this income class the rate of in- very eloquently how deeply deductions have eroded tax- crease of tax burden is in 20-40% range. The next largest able income in Japan. Table 4 (B) is a counterpart of Table portion of people in the same income class will see a more 3 (B) and shows the variance of the effects of cutting de- than 80% increase of their tax. This is due to the fact that ductions on tax collection, where the rate of deduction cut they cut their taxable income very significantly and the is set at 30%. Revenue-increasing effects of cutting de- elimination of deductions pushes up their tax liabilities ductions are on average larger for lower income brackets, (probably by letting their taxable income move up to a

Figure 1: Comparison of the current and linear-income tax systems Tax

45-degree line

A D

B b=0

O Y1 Y0 Y2 Income before tax b=b1 C

Note: The 45-degree line is the maximum tax that can be collected from income. The bold curve OA rep resents the tax collected under the current Japanese tax system. When the basic transfer, b, is set equal to zero, the linear-income tax is shown as the line OB. The slope of this curve is the marginal tax rate. When b=b1 (>0), the linear-income tax is CD with OC set equal to b1 and the marginal tax becomes greater than that under b=0. When the current sy stem is rep laced by the linear-income tax with b=0, peop le with income below Y0 lose and above gain. Whereas when b=b1, people with income between Y1 and Y2 lose and with either below Y1 or above Y2 gain.

S 22 higher bracket). Since b is the transfer that is distributed to all households, it reduces the tax liability across the board. The marginal tax rate is uniform and applied to every income without 4. Broadening the Tax Base by Linear deductions (in the purest form of the linear income tax). Income Tax Figure 1 compares the current Japanese and the linear income tax systems. Income before tax, i.e., without 4.1 How the Linear Income Tax Works deductions, is on the horizontal axis and the tax liability on The preceding two sections have shown that the Japa- the vertical. The 45-degree line is drawn in the figure as a nese personal income tax has shifted its burden to taxpay- reference indicating the maximum that can be taxed given ers in higher brackets: in fact, to the top bracket in our the income. The tax collected under the current Japanese decile estimate. The reason we have found for this is that tax system is shown by the curve OA: various deductions slashing their taxable income cuts the tax burden of lower make tax liabilities zero up to a certain level of income and income people. In fact, the amount of each component of people start paying tax above the threshold with increas- deductions is fixed per person; hence, the proportion of ing marginal tax rate. The steeper slope means a higher income that people in lower brackets can deduct is much marginal tax rate. larger than those in higher brackets. But given that the When switching to the linear income tax, the mar- personal income tax is one of the most important sources ginal tax rate is set uniform. Hence, the tax curve now be- of revenue in Japan, the revenue lost in granting lenient comes a straight line in the figure. The amount of the basic deductions must be recouped by some other measures. transfer appears in the figure as the intercept of the line on Increasing the marginal tax rate is an answer to this and we the vertical axis. We assume a revenue-neutral have found that the rate jumps at the top bracket. and compare the current Japanese tax with the linear in- Now looking back on the Japanese personal in- come tax. This implies that when the basic transfer is set come tax, we think that the capacity of the tax to address higher, the marginal tax rate has to be set higher as well. In the issue of equity has been too restricted. An insight that the experiments that follow, we determine the amount of has been missing is that income distributional problems the basic transfer first, and calculate the marginal tax rate can be dealt with within the tax system, that is to say, by that satisfies the revenue constraint. giving transfer either across the board or to certain se- When the basic transfer, b, is set equal to zero, the lected groups. In this expanded setup of the personal in- linear income tax is shown as the line OB. The marginal tax come tax, an equity objective of the tax can basically be rate, t, is the slope of the line. When the basic transfer is increased and set at b (>0), the linear income tax becomes assigned to transfers, i.e., tax credits in the jargon of per- 1 the line CD with OC equal to b and the marginal tax be- sonal income tax, and the marginal tax rate should be set as 1 flat as possible to minimize the efficiency costs of the tax. comes greater than under the case where b is set at zero. A simple personal income tax system that meets The figure also shows who gains and loses by the both the goals of equity and efficiency is a linear income reform. When the current system is replaced by the linear tax (Mirrlees 1971, 1997; Stern 1976; Atkinson, 1995; Tajika income tax without the basic transfer, people with income below Y lose and those with income above Y gain. Since and Furutani 2000). This tax system consists of a basic 0 0 transfer which is given to all individuals and a uniform large deductions of the current tax system are dispensed, marginal tax rate. In a sense (though not perfectly correct), people in lower brackets of income have to pay more tax the basic transfer corresponds to the equity and the uni- than under the current system: however, since the mar- form marginal tax rate to the efficiency goal. Put simply, the ginal tax rate is cut significantly, those who are in the higher higher basic transfer can better cope with equity problems income brackets will gain. In the figure the threshold in- come is Y and the uniform and low marginal rate can avoid disincen- 0. When the basic transfer is increased and set at b , tives of working. 1 We now turn to an outline of the mechanics of the the marginal tax rate increases to satisfy the revenue con- straint. It turns out that people with income below Y and linear income tax. Denoting the basic transfer b and the 1 above Y gain and those in between lose. Under the linear marginal tax rate in the linear income tax system t, we can 2 express the tax liability, T, for income Y as: income tax every individual receives the transfer and owes T= - b + tY. tax simultaneously. But people in lower brackets will gain

S 23 in net, for their tax liability is smaller than the transfer re- personal income tax. ceived. At the same time, so long as the marginal tax rate Table 5 (A) shows the case where the basic trans- does not climb up exorbitantly to meet the required rev- fer is zero, i.e., b=0. Since there is no basic transfer and the enue, people with income above a certain level gain, too. It reform must be revenue neutral, the average and marginal may be worth noting that the threshold of income above tax rates become equal and they are also identical with the which people gain under the linear income tax increases as overall average tax rate under the current system. The av- the basic transfer increases, i.e., Y0 to Y2 in the figure. erage as well as the marginal tax rate thus calculated turns out to be 0.047. 4.2 Effects of the Reform on Average and Marginal Tax The average tax rate for lower income classes in- Rates creases significantly due to the reform. Even in income With the conceptual background of the reform discussed class 4 it increases from the current 0.028 to 0.047. The above, we would like to show its effects on average and marginal tax rate also increases due to the reform up to marginal tax rates of each income class. Table 5 reports the income class 7, but since our estimate of the marginal rate effects and the average amount of gains or losses for each of Japanese tax system is not increasing in income (there class. The sample we use here is the same as the one in are declines in lower income classes), the results are not section II in which we examined the progressivity of the straightforward.

Table 5: Average and marginal tax rates under the linear-income tax (A)Case1: b=0 The current Japanese Linear-Income Tax system Income Average Income Average Marginal Average Marginal Average Gains Class JPY1,000 Tax Rate Tax Rate Tax Rate Tax Rate JPY 1,000 1 669 0.003 - 0.047 0.047 -29.5 2 1,253 0.013 0.025 0.047 0.047 -42.7 3 1,972 0.025 0.045 0.047 0.047 -44.5 4 2,625 0.028 0.039 0.047 0.047 -49.7 5 3,271 0.032 0.047 0.047 0.047 -50.2 6 4,011 0.033 0.039 0.047 0.047 -56.2 7 4,874 0.035 0.041 0.047 0.047 -61.3 8 5,979 0.038 0.054 0.047 0.047 -54.2 9 7,518 0.047 0.080 0.047 0.047 -4.3 10 11,888 0.080 0.138 0.047 0.047 392.0 Overall 4,407 0.047 - 0.047 - 0

(B)Case 2: b=20% and 50% of the average tax paid under the current system The current Japanese Linear-income tax with Linear-income tax with system b=20% of average tax b=50% of average tax Income Average Income Average Marginal Average Marginal Average Gains Average Marginal Average Gains Class JPY1,000 Tax Rate Tax Rate Tax Rate Tax Rate JPY 1,000 Tax Rate Tax Rate JPY 1,000 1 669 0.003 - -0.006 0.057 5.9 -0.085 0.071 59.0 2 1,253 0.013 0.025 0.023 0.057 -12.9 -0.012 0.071 31.9 3 1,972 0.025 0.045 0.036 0.057 -21.5 0.018 0.071 13.1 4 2,625 0.028 0.039 0.041 0.057 -32.8 0.031 0.071 -7.5 5 3,271 0.032 0.047 0.044 0.057 -39.4 0.039 0.071 -23.3 6 4,011 0.033 0.039 0.046 0.057 -52.4 0.045 0.071 -46.8 7 4,874 0.035 0.041 0.048 0.057 -65.7 0.050 0.071 -72.3 8 5,979 0.038 0.054 0.050 0.057 -69.1 0.054 0.071 -91.4 9 7,518 0.047 0.080 0.051 0.057 -33.8 0.057 0.071 -77.9 10 11,888 0.080 0.138 0.053 0.057 321.1 0.062 0.071 214.9 Overall 4,407 0.047 - 0.047 - 0 0.047 0.071 0.0 Note: The basic transfer under the linear-income tax, b, is set at 20% and 50% of the average tax under the current personal income tax of Japan

S 24 On the other hand, the reform brings forth reduc- tion of the marginal tax rate of the top income class from tion of the marginal tax rate for income classes 8 and above, 0.138 to 0.057 when the basic transfer is 20% of the aver- and the average tax burden is also cut for the top bracket. age tax (and to 0.071 when it is 50%) is still a very remark- It is amazing to see that the marginal tax rate of the top able change. income class is reduced from 0.138 to 0.047 by switching Of course, not everybody is better off in this zero- to the linear income tax. sum game: the people in the middle lose. Again, this is a Thus, the reform without a basic transfer shifts bur- matter of equity and efficiency. We should note that the den from higher to lower income classes. This is clearly experiments we have done imply that the personal income seen in the last column of Table 5 (A), which reports the tax with distribution has a chance to reduce the marginal average gains for each income class. Two remarks are due tax rate significantly. If there is a strong case in Japan for here. The first is that it is because the tax burden is so cutting the marginal tax rate for the top bracket or brackets progressive under the current system, only the top in- in its vicinity, our study suggests a way for such change. come class gains on average by the reform: under a more Besides we can improve our policy by tailoring transfers moderate tax schedule, people in more higher income more carefully so that they can suit the needs of those classes would gain. This is what we can read from Figure 1 who really deserve them. above. The second is that a big reduction in the marginal tax rate occurs due to the reform. In view of the fact that its economy has been experiencing a decade long decline, 5.Conclusions the efficiency matters more than at any time before in Ja- The persisting view of taxation amongst policy makers in pan. Hence, the effect of the reform on the marginal tax rate Japan has been that tax is a means for collection and that it merits attention. should not distribute within itself. Given strong pressures We turn to Table 5 (B). Here, the basic transfer is for reducing tax rate and for preferential treatments from raised to 20% and 50% of the average tax under the current virtually every corner of the street, this guarded attitude personal income tax, which is about JPY207,000. When the of policy makers toward taxation may be understandable. basic transfer is set at 20% and 50% of the average tax, the Nevertheless, the costs of the dichotomy in tax marginal tax rates under the linear income tax that satisfy policy seem to be enormous. Since granting a basic trans- the revenue constraint are 0.057 and 0.071, which are higher fer is set aside from the role of taxation and therefore very than the rate where the basic transfer is zero, i.e., 0.047. few tax credits are institutionalized, the personal income tax has to satisfy the progressivity of tax burden for itself. The effects and implications of the reform on vari- This was done mainly by introducing various deductions, ous income classes are as follows. When the basic trans- and the taxable income was squeezed substantially. Since fer increases, people in lower income classes start to gain most of the items of deductions are fixed per person, those and those in upper income classes start to lose: when the with smaller income enjoy better treatment when paying basic transfer is 20% of the average tax, income class 1 tax. gains, and when it is 50%, income classes 1 through 3 The paper has shown that lost revenue is recov- gain. The top class continues to gain when compared with ered by increasing marginal tax rate of the people in higher the present tax system, but the magnitude by which they income brackets. The magnitude and variance of deduc- gain declines. tions are then discussed. As a policy option for reducing Since the tax burden is mitigated by the basic trans- the marginal tax rate, we have explored the case for com- fer, the average tax rate is not so high as under the case bining collection and distribution within the personal in- without transfer. Income class 4 now faces much more mod- come tax; as a rudimental experiment of such policy, the erate increases in their average tax burden: when the basic effects of switching from the current tax to the linear in- transfer is 20% and 50% of the average tax, their average come tax systems are examined and a significant chance tax rates are 0.041 and 0.031 respectively, which are smaller for reducing the marginal tax rate is observed. than 0.047 under the reform without transfer. Hence, the Several issues remain untouched but must yet be results of the reform look persuasive. As for the effects on explored. First, the paper has dealt only with taxation of the marginal tax rate, they are not as significant as in the earned income. Taxation of other kinds of income, espe- case without the basic transfer. Nevertheless, the reduc- cially the incomes of the self-employed and pension re-

S 25 ceivers, deserves to be studied. Second, when distribu- income taxation.” Review of Economic Studies, 38:175- tion is combined with collection in our simulation, the ba- 208. sic transfer is distributed uniformly to every individual Stern, Nicholas H., 1976, “On the specification of optimum equally. This clearly is too expensive a policy. As in the income taxation.” Journal of Public Economics, 6: 123- cases of earned-income and working-family credits in the 62. United States, the United Kingdom, and other countries, Tajika, Eiji and Izumi Furutani, 2000, “Japanese personal more thoughtful distribution of the transfer from within income tax: facts and the theory.” (in Japanese), Finan- the tax system has to be studied. Last, incentive aspects cial Review, 53: 129-161. of the personal income tax are intentionally dismissed in the paper. While this may be acceptable for showing the evidence of a tax policy, we would not be able to bypass them when designing appropriate transfers. Eiji Tajika (Graduate School of Economics,Hitotsubashi Univer- sity) Notes Izumi Furutani * This paper is an outcome of research for a project, Distri- (Policy Research Institute, Ministry of ) bution of Income, which is a sub-project of Kosei Kagaku Kenkyu Hojokin Jigyo “International Coopera- tion Project on Reforms of Social Security (1999-2001).” And the survey used in the paper was made available to us by the Ministry of Health, Labor and Welfare of Japan (SID No.117 dated 3rd April 2001). We are very grateful for discussions and comments of the members of our research group. The views expressed here are entirely ours and do not represent those of the organi- zations to which we are affiliated. Needless to say, we are responsible for any remaining error. 1 There are some exceptions, but they are for delivering incentives. Tax credits for owner occupied housing and investment tax credits for small-size businesses are typi- cal of those incentive-motivated policies.

References Atkinson, A. B., 1995, Public economics in action: the basic income/ proposal. Clarendon Press, Oxford. Ministry of Finance of Japan, 2000, An outline of Japa- nese taxes. The Government Printing Bureau. (http://www.mof.go.jp/english/tax/tax.htm) Ministry of Health, Labor and Welfare of Japan, (http://www.mhlw.go.jp/english/index.html#statistics) Mirrlees, James A., 1997, “Information and incentives: the economics of carrots and sticks.” The Economic Journal, 107:1311-1329. , 1971, “An exploration in the theory of optimum

S 26